|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: JDH DJS. Performed the experiments: JS JDH. Analyzed the data: JS DJS. Wrote the paper: JS JDH DJS.
Only a portion of the estimated heritability of breast cancer susceptibility has been explained by individual loci. Comparative genetic approaches that first use an experimental organism to map susceptibility QTLs are unbiased methods to identify human orthologs to target in human population-based genetic association studies. Here, overlapping rat mammary carcinoma susceptibility (Mcs) predicted QTLs, Mcs6 and Mcs2, were physically confirmed and mapped to identify the human orthologous region. To physically confirm Mcs6 and Mcs2, congenic lines were established using the Wistar-Furth (WF) rat strain, which is susceptible to developing mammary carcinomas, as the recipient (genetic background) and either Wistar-Kyoto (WKy, Mcs6) or Copenhagen (COP, Mcs2), which are resistant, as donor strains. By comparing Mcs phenotypes of WF.WKy congenic lines with distinct segments of WKy chromosome 7 we physically confirmed and mapped Mcs6 to ~33 Mb between markers D7Rat171 and gUwm64-3. The predicted Mcs2 QTL was also physically confirmed using segments of COP chromosome 7 introgressed into a susceptible WF background. The Mcs6 and Mcs2 overlapping genomic regions contain multiple annotated genes, but none have a clear or well established link to breast cancer susceptibility. Igf1 and Socs2 are two of multiple potential candidate genes in Mcs6. The human genomic region orthologous to rat Mcs6 is on chromosome 12 from base positions 71,270,266 to 105,502,699. This region has not shown a genome-wide significant association to breast cancer risk in pun studies of breast cancer susceptibility.
Breast cancer susceptibility is a complex trait controlled by genetic, epigenetic, and environmental components. Susceptibility to breast cancer is comprised of low, moderate, and high penetrance risk alleles. Linkage analysis in high incidence families and population-based genetic association studies using cases and controls have identified breast cancer risk alleles; however, a large portion of the estimated heritability in breast cancer susceptibility remains to be explained , . Laboratory rats (Rattus norvegicus) have been used to identify human breast cancer risk alleles using comparative genetics . To identify loci for comparative studies, rat strains that exhibited phenotypic variation in susceptibility to chemical and hormone induced mammary carcinogenesis have been used to predict mammary carcinoma susceptibility (Mcs) and estrogen-induced mammary cancer QTLs . Rats were chosen to identify Mcs QTLs because chemically induced rat mammary carcinomas have similar histopathology to human breast carcinomas . The Wistar Furth (WF) rat strain is susceptible to developing mammary carcinomas, while the Wistar Kyoto (WKy) and Copenhagen (COP) rat strains are resistant , . A QTL linkage analysis of rat Mcs in females from [WF x WKY]F1 x WF backcrosses predicted four Mcs QTLs named Mcs5, Mcs6, Mcs7, and Mcs8 . The Mcs6 QTL displayed a peak linkage at marker D7Mit28 and significant linkage was found between markers D7Mit28 and D7Rat45. The predicted Mcs6 QTL region overlapped with Mcs2, a Mcs QTL predicted using the mammary cancer resistant Copenhagen (COP) rat strain . The peak LOD score for the predicted Mcs2 QTL was at genetic marker D7Uwm11. Here, we physically confirm both Mcs2 and Mcs6 QTLs, and map Mcs6 to an approximately 33 Mb region of rat chromosome 7, which is orthologous to a region on human chromosome 12.
Figure 1 depicts the WF.WKy congenic lines used to physically confirm and map the predicted Mcs6 QTL to a location on rat chromosome 7 between markers D7Rat171 and gUwm64-3. Table 1 summarizes mammary carcinoma susceptibility (Mcs) phenotypes from lines shown in Figure 1. Respectively, WF.WKy lines A, B, and F had mammary carcinoma multiplicity phenotypes (mean ± SD carcinomas per rat) of 3.5±2.5, 3.9±2.6, and 3.4±2.1. These lines developed significantly less mammary carcinomas compared to susceptible WF control animals, which had a mean ± SD of 7.3±3.6 carcinomas per rat (Mann-Whitney test post hoc p-values=0.0012, 0.0007, and 0.0001). Females from WF.WKy lines A, B, and F had similar mammary carcinoma multiplicity phenotypes when compared to each other (Kruskal-Wallis test p-value=0.8171). WF.WKy line C had a mean ± SD of 8.0±4.1 mammary carcinomas per rat, and line E had a mean ± SD of 7.2±3.0. Neither line C nor line E was significantly different from WF control animals (Mann-Whitney test post hoc p-values=0.7521 and 0.7893, respectively). Decreased susceptibility WF.WKy lines A, B, and F overlapped maximally from markers D7Rat171 to D7Rat45, and minimally from gUwm64-24 to gUwm64-3. We conclude that these lines physically confirm Mcs6 as an independently acting Mcs QTL.
WF.WKy line C, which was not significantly different from susceptible WF control animals, overlapped the distal end of the Mcs6 region demarked by lines A, B, and F from markers gUwm64-3 to D7Rat128. This ruled out that this region contained Mcs6 or independently acting susceptibility QTLs. The susceptible phenotype of line E, which partially overlapped susceptible line C, confirmed that the region from gUwm64-12 to D7Rat128 did not contain an independently acting Mcs QTL. When considered together, our results from WF.WKy congenic lines delimited Mcs6 to a 33 Mb region of rat chromosome 7 from markers D7Rat171 to gUwm64-3.
Mcs2 is a Copenhagen (COP) rat Mcs QTL that was also predicted to reside on chromosome 7 . The original QTL scans of Mcs2 (COP) and Mcs6 (WKy) predicted regions of significant linkage on rat chromosome 7 that overlapped from markers D7Mgh15 to D7Mgh10 . We physically confirmed Mcs2 using WF.COP congenic lines A and B that each contained a COP segment of rat chromosome 7 from markers D7rat39 to D7Uwm12 (Figure 1). These markers corresponded to rat chromosome 7 base positions 4,936,704-86,028,057. WF.COP lines A and B had similar mammary carcinoma multiplicity phenotypes of 2.4±1.6 and 2.0±1.5 mean ± SD carcinomas per rat (Table 1, Mann-Whitney test p-value 0.4510). Both lines were significantly less than the WF susceptible phenotype (Mann-Whitney post hoc test p-values <0.0001). The confirmed Mcs2 QTL overlapped the delimited Mcs6 QTL between markers D7Rat171 and gUwm64-3 (Figure 1). The pooled phenotypes of Mcs2 (WF.COP lines A and B) and Mcs6 (WF.WKy lines A, B, and F) females were, respectively, 2.2±1.5 and 3.6±2.4 mean ± SD mammary carcinomas per rat. These phenotypic values were statistically different (Mann-Whitney test p-value=0.0030) suggesting that the Mcs2 COP allele may have additional genetic variation contributing to mammary carcinoma resistance.
Using the UCSC rat and human genome browsers (genome.ucsc.edu) we determined that the Mcs6 QTL from markers D7Rat171 to gUwm64-3 (chr7:22,382,725-55,364,398) aligns to a contiguous 34.2 Mb region of human chromosome 12 from base positions 71,299,117 to 105,502,699, which was inverted with respect to the rat genome assembly. Within these orthologous regions 37.7% of the bases over 99.9% of the span were identical. Rat genomic sequence flanking Mcs6 aligned to human chromosome 12 up to rat chromosome 7 base positions 22,310,240 on one flank and 55,442,351 on the other flank. Moving out past these syntenic regions of rat chromosome 7 and human chromosome 12 resulted in rat genomic sequence that aligned to human chromosome 16 on one flank and chromosome X on the other.
Figure 2 shows annotated genes in the Mcs6 QTL between markers D7Rat171 and gUwm64-3. These genes were on the UCSC rat November 2004 assembly (build 3.4/rn4). Not including pseudogenes, this region contained 111 annotated genes including two microRNAs (Mir135 and Mir331). There were 137 annotated genes found in Mcs6 using the Rat Genome Database (rgd.mcw.edu), and 215 found using the Ensembl Genome Browser (uswest.ensembl.org) rat database. Independent searches of Mcs6 potential candidate genes using gene ontology terms limited to mammary, hormone, estradiol, or DNA repair resulted in the following gene list: Socs2, Igf1, Nr1h4, Osbpl8, Nr2c1, Pmch, Stab2, Nts, Lgr5, Btg1, Lta4h, and Ube2n. All genes located in rat Mcs6 were also annotated in the human orthologous region. There were a total of 243 genes listed in the human orthologous region when the UCSC human genome browser was queried and 226 genes when the Ensembl browser was used.
We queried the Oncomine 4.4 database (www.oncomine.org, Compendia Bioscience) to identify Mcs6 human orthologs that may have been reported as differentially expressed between non-diseased breast tissue and ductal breast carcinomas. A list of the studies included can be found in References S1 (S1-S7). Results from this search are presented in Table S1. There were twenty-nine transcripts reported with p-values <0.05, and six of these had p-values <0.001 Mcs6 human orthologs with p-values <0.001 were ALX1, DUSP6, EPYC, PPP1R12A, TXNRD1, and ZDHHC17.
To determine if genetic variation in the Mcs6 orthologous human genomic region from chromosome 12:71,270,266-105,502,699 might be associated with breast cancer susceptibility we searched the NCBI dbGAP database using the association results browser set to “breast neoplasms” and chromosome 12 (www.ncbi.nlm.nih.gov/gap). We also searched published breast cancer risk GWA studies for positive associations in this region. A list of the association studies included can be found in References S1 (S8-S23). While five human SNPs were identified that potentially associated with breast cancer susceptibility, tested genetic variants did not reach a genome-wide significance level of P< 10-7 proposed by Thomas et al. . A summary of human SNPs in the Mcs6 orthologous region that potentially associate with breast cancer risk can be found in Table S2.
Both mammary cancer susceptibility QTL mapping in rats and low-penetrance breast cancer risk associated allele identification in humans are in early stages of discovery. Results of human population-based and rat genetic studies completed to date indicate that a majority of breast cancer risk alleles are low-penetrance, and pathways in addition to DNA repair and hormonal regulation control susceptibility. Furthermore, a majority of the risk associated polymorphisms that have been identified reside in non-protein coding DNA. Given the commonalities of low-penetrance, novel pathways, and non-protein-coding risk alleles, it is possible that there is some degree of overlap in susceptibility loci between these two organisms. There may be shared susceptibility loci that can be more easily identified in rats than by current human GWAS methods. If this is the case, resolving rat mammary carcinoma susceptibility QTLs to single loci so that the human orthologs can be extensively tested in human population-based genome-targeted association studies may continue to be fruitful. This was the case for Gould and colleagues who used rat Mcs5a to successfully target the orthologous human genomic region to identify two breast cancer susceptibility alleles on human chromosome 9 . Quan et al. have shown that rodent susceptibility QTLs identified for a specific tumor type may also be useful to identify common susceptibility alleles for other tumor types . Thus, the amount of overlap between rodent and human common disease susceptibility loci may be appreciable.
Here, predicted rat mammary carcinoma susceptibility QTLs Mcs6 and Mcs2 were physically confirmed using congenics. In our study Mcs2 and Mcs6, respectively, reduced the mammary carcinoma susceptibility phenotype of the WF strain by 70 and 50% when homozygous. In the original QTL studies, one Mcs2 COP or one Mcs6 WKy allele was predicted to reduce tumor multiplicity by 43% or 40%, respectively , . While those studies were not completed using congenics, and we have not tested heterozygous females, comparing the results of tumor numbers of homozygous Mcs2 and Mcs6 congenics with heterozygous females from the previous studies suggests there may be some degree of dominance from the WF allele at this locus. However, these differences may also be explained by background COP and WKy alleles that would have been present in females used for the QTL linkage studies. The stronger effect of the Mcs2 COP allele compared to the Mcs6 WKy allele may indicate that the Copenhagen strain has additional genetic variation within the region that contributes to reducing the Mcs phenotype.
In this study Mcs6 was mapped to a 33 Mb region from markers D7Rat171 to gUwm64-3, which corresponds to rat chromosome 7 base positions 22,382,725-55,364,398. The peak LOD scores for genetic markers in the predicted Mcs6 and Mcs2 QTLs were at D7Mit28 and D7Uwm11, respectively , . Both peak QTL markers fall within the narrowed interval confirmed here with congenics. D7Mit28 is at base position Chr7:29,204,226, and D7Uwm11 is at base position Chr7:50,160,469. This supports using QTL linkage analysis of rat complex phenotypes to identify regions of the genome on which to focus targeted human genome association studies.
Alignment of the rat Mcs6 QTL region to the human genome reference sequence revealed that the human orthologous region is entirely positioned on 34.2 Mb of chromosome 12, and is inverted compared to rat Mcs6. To our knowledge no polymorphisms in the human genome region orthologous to rat Mcs6 have shown a genome-wide significant association to breast cancer risk in a GWAS. An intriguing potential candidate gene in Mcs6 is insulin-like growth factor 1 (Igf1). Human candidate-gene association studies that have tested IGF1 locus polymorphisms reported mixed results ranging from significant low-penetrance to null associations between IGF1 alleles and breast cancer risk –. Increased IGF1 levels have been associated with breast cancer and increased breast cancer susceptibility , . Common genetic variation in IGF1 has been associated with circulating levels of IGF1 in a population-based study . Other Mcs6 potential candidates have human orthologs that have been associated with cancer or other human diseases. SLC5A8 and GLIPR1 are predicted tumor repressor genes , . METAP2 is predicted to be anti-angiogenic . A human genome-wide linkage analysis of posterior amorphous corneal dystrophy identified a region of human chromosome 12 that is orthologous to a Mcs6 region that contains genes DCN, LUM, KERA, and EPYC . PCTK2 has been shown to be upregulated in sporadic breast cancer biopsies in women of Mexican ancestry . Suppressor of cytokine signaling 2 (Socs2) is also a promising potential candidate gene in Mcs6. Higher SOCS2 levels have been associated with better breast cancer prognosis –. Socs2 is an important mediator of prolactin regulated mammary gland development in the mouse . SOCS2 is upregulated by 17β-estradiol in cancer cell lines , and SOCS2 expression was found to be higher in breast carcinomas compared to non-diseased breast tissue .
Mcs6 human orthologous transcripts that have been reported as differentially expressed between non-diseased breast tissue and ductal carcinoma tissue encode for two phosphatases (DUSP6, PPP1R12A), a proteoglycan (EPYC), a redox regulator (TXNRD1), and two uncharacterized proteins (ALX1, ZDHHC17). These findings suggest that the rat Mcs2 and Mcs6 QTLs control potentially novel mechanisms contributing to mammary cancer susceptibility that could translate to breast cancer.
Human genetic variation in the Mcs6 orthologous genomic region has been reported to be associated with breast cancer susceptibility, but tested polymorphisms have not reached genome-wide level significance. This highlights the importance of this region to breast cancer genetics. It suggests, as many have, that Bonferroni correction based methods may be too stringent to control for multiple comparison testing in human population-based association studies. The confirmed rat Mcs2 and Mcs6 QTLs support that some or all of the potential human breast cancer risk associations are true positives and warrant further investigation.
We conclude that the human genomic region on chromosome 12 from base positions 71,270,266 - 105,502,699 is orthologous to rat Mcs6. This region may be a good target for extensive population-based genetic association studies to determine if humans have a concordant breast cancer susceptibility allele. Rat Mcs6 and Mcs2 congenic lines will be useful to study breast cancer genetic susceptibility mechanisms.
Congenic lines were established using previously described methods  and are defined as rats with specified WKy chromosome 7 segments from the predicted Mcs6 QTL or the specified COP chromosome 7 segments from the predicted Mcs2 QTL introgressed into a WF genetic background. Rats were housed in an AALACC approved facility and protocols were approved by a University of Louisville institutional animal care and use committee under protocol number 07118. Genotyping was performed at the markers shown in Figure 1 according to methods previously described . Primer and sequence information for most genetic markers used is available at the Rat Genome Database (rgd.mcw.edu) . The following primer sequence pairs are for genetic markers that were submitted to the NCBI probe database, but may not yet be publicly available: gUwm64-24, 5′-TGTCCTCTCATACCACTTCC-3′, 5′-CGGTGGTCATACTCTGAAT-3′; gUwm64-3, 5′-AACAGTTCTCCTTTCCCTTC-3′, 5′-TCTCTTTGCCAGTCTGTTTT-3′, gUwm64-12, 5′-CTGGTCATTCAACAGGAAAT-3′, 5′-CCTATCCCTGTTCTTCCTCT-3′; and gUwm64-19, 5′-AACAATAGGCACAGTGAT-3′, 5′-ATGCTGTCAGAACGGTAAGT-3′.
Female congenic and WF (Harlan) rats at 50-55 days of age were treated with a single dose of 7,12-Dimethylbenz(a)anthracene (DMBA, ACROS Organics, Fisher Scientific) in sesame oil by gastric intubation at 65 mg DMBA/kg body mass. Mammary carcinomas ≥3×3 mm2 were counted at 15 weeks following DMBA treatment. Mammary carcinoma multiplicity data were analyzed to compare congenic line phenotypes to susceptible WF control females. WF.WKy chromosome 7 congenic phenotypes for each line were from the following generations: line A, N11F2 and N11F4 (total n=17); line B, N11F2 and N11F3 (total n=32); line C, N14F2, N14F3, and N14F4 (total n=23); line E, N14F2 and N14F4 (total n=43); and line F, N14F3 and N14F4 (total n=28). WF.COP chromosome 7 congenic phenotypes were from generations: line A, N10F5 (n=16); and line B, N11F3 and N11F4 (total n=16).
To compare mammary carcinoma susceptibility phenotypes, mammary carcinoma multiplicity data were analyzed using Statview (SAS Institute). Mammary carcinoma multiplicity phenotypes for more than two groups were compared using the Kruskal-Wallis nonparametric test. If this test was statistically significant (p value <0.05) multiple Mann-Whitney nonparametric tests were performed by comparing each congenic line phenotype to the susceptible WF control group. To correct for multiple comparisons a p-value ≤0.007 (α=0.05/7 comparisons) was required for statistical significance. For analysis of two groups, Mann-Whitney nonparametric tests were performed and p-values ≤0.05 were required for statistical significance.
Identification of the human orthologous region and transcripts mapping to human and rat regions of interest were performed using the RGD, UCSC, and Ensembl genome browsers -. The genome assemblies used were Homo sapiens version GRCh37/hg19 and Rattus norvegicus version 3.4/rn4. The percent sequence identity between human and rat was determined using the “Convert” option at the UCSC Genome Browser. The gene ontology search was completed by downloading from the Rat Genome Database gene ontology terms for genes annotated in Mcs6 and searching this database by terms: “mammary”, “hormone”, “estradiol”, or “DNA repair”.
Results of a query using Oncomine to identify human Mcs6 orthologous transcripts that have been shown to be differentially expressed between non-diseased and ductal carcinoma breast tissues.
Human polymorphisms located in the Mcs6 orthologous region that have not reached genome-wide significance, but are potentially associated with breast cancer susceptibility.
Citations for supplemental references.
The authors thank Aaron Puckett, Cordes Cooper, and Paula Ginter for their assistance with phenotyping and genotyping animals used in this study.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This work was supported by National Institutes of Health grants CA137052, CA077494, and ES014443. J.S. was supported by the University of Louisville Integrated Programs in Biomedical Sciences. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.